Benchmarking atlas-level data integration in single-cell genomics - integration task datasets

Pancreas, Lung atlas, human immune cell, and human and mouse immune cell integration tasks from the manuscript "Benchmarking atlas-level data integration in single-cell genomics". <br><br>These datasets were aggregated from public datasets, cell annotations were harmonized or reannotated, and the data was consistently preprocessed using scran pooling and log+1 transformation. In the immune cell datasets an erythrocyte development trajectory was also annotated. Details on dataset preprocessing can be found in the paper and in the accompanying Github at https://www.github.com/theislab/scib.<br><br>Please cite the paper and the papers the individual datasets were aggregated from when using this data.<br><br>Note: the "complexBatch" version of the pancreas data was used in the paper. This version contains a higher resolution batch covariate and all batches were normalized jointly.<br>